Skip to main content

Skills application

Instruction and application
In Progress

Time-series features for energy prediction​

In this activity, you will apply feature engineering techniques to an energy consumption dataset to improve demand forecasting. You’ll generate time-based features, incorporate domain-specific signals and optimise feature selection to enhance model accuracy.

These skills are essential for building predictive models that drive data-informed decision-making, helping businesses to improve efficiency, allocate resources effectively and adapt to changing patterns in their industry.

Challenge instructions and resources​

To complete this activity, you’ll need both the Jupyter Notebook and the Energy Consumption datasets. Use the links below to access and download the required files:

mv-l6-m4-w2.ipynbWorld_Energy_By_Country_And_Region_1965_to_2023.csv

Activity context​

As a Data Scientist for a city’s energy provider, you are tasked with improving electricity demand forecasting. The company collects hourly consumption data across districts, aiming to predict next-day power demand to optimise distribution. Using the provided historical consumption dataset, your goal is to engineer time-series features that enhance prediction accuracy and support efficient energy management.

Work on the challenges​

Follow the instructions in the Jupyter Notebook to apply feature engineering techniques to the energy consumption dataset, including creating time-based features, incorporating external factors and optimising feature selection.

Collaborate in the breakout room​

Discuss and share insights with fellow apprentices as you work on the activity challenges. While the activity is designed for individual completion, feel free to ask questions, collaborate and compare approaches with your group.

Regroup and share​

Return to the main session after 30 minutes to discuss key takeaways and insights from the exercise.

Action item: Activity share out​

Which time-based features had the most impact on forecast accuracy, and why do you think they were effective?